Thermal atomic layer etching of cobalt using sulfuryl chloride for chlorination and tetramethylethylenediamine or trimethylphosphine for ligand addition

Thermal atomic layer etching (ALE) of cobalt was developed using sulfuryl chloride (SO2Cl2) for chlorination and either tetramethylethylenediamine (TMEDA) or trimethylphosphine (PMe3) for ligand addition. In situ quartz crystal microbalance (QCM) measurements were used to monitor the thermal ALE of cobalt using the SO2Cl2/TMEDA and SO2Cl2/PMe3 processes. For every SO2Cl2 exposure, there was a mass gain during chlorination. For every TMEDA or PMe3 exposure, there was a mass loss during ligand addition. The result was a net removal of cobalt during each chlorination/ligand-addition reaction cycle. Average etch rates determined from QCM measurements for the SO2Cl2/TMEDA process at 175, 200, 225, 250, 275, and 300 degrees C were 0.62 +/- 0.41, 1.35 +/- 0.64, 2.31 +/- 0.91, 6.43 +/- 1.31, 10.56 +/- 2.94, and 7.62 +/- 4.87 A/cycle, respectively. These etch rates were corroborated using x-ray reflectivity (XRR) studies on cobalt thin films on silicon coupons. Quadrupole mass spectroscopy analysis also revealed that the cobalt etch product from TMEDA exposures on CoCl2 powder was CoCl2(TMEDA). The SO2Cl2/TMEDA process could remove the surface chloride layer formed by each SO2Cl2 exposure with one TMEDA exposure. In contrast, the SO2Cl2/PMe3 process required 20-40 individual PMe3 exposures to remove the surface chloride layer formed from each SO2Cl2 exposure at 130-200 degrees C. An increasing number of PMe3 exposures were needed as the temperature decreased below 130 degrees C. The etch rates for the SO2Cl2/PMe3 process with multiple PMe3 exposures were 2-4 A/cycle as determined by the QCM and XRR studies. For both the SO2Cl2/TMEDA and SO2Cl2/PMe3 processes, the etch rate was determined by the amount of CoCl2 created during the SO2Cl2 exposure. Thicker surface CoCl2 layers from larger SO2Cl2 exposures resulted in higher Co etch rates that could exceed one crystalline unit cell length. Atomic force microscopy measurements determined that the cobalt surface roughness decreased after Co ALE with the SO2Cl2/TMEDA process. In contrast, the cobalt surface roughness increased after Co ALE with the SO2Cl2/PMe3 process. The chlorination and ligand-addition mechanism should be generally applicable for metal ALE for metals that form stable chlorides.